Recruiting is difficult. Training quickly those you already have is less so.

In 2026, 85% of French industrial companies report recruitment difficulties. According to the 2026 Workforce Needs Survey (BMO) from France Travail, 211,000 industrial positions are to be filled this year, of which 48% are considered difficult to satisfy. Other surveys estimate the number of vacant positions in the industry during the same period at 130,000. This skills shortage primarily affects the most qualified technical jobs, and it will not be resolved with another recruitment campaign.

A shortage concentrated on the jobs that keep the workshop running

Certain positions concentrate the bulk of the tension. According to Techniques de l'Ingénieur, two profiles dominate the difficult recruitments in 2025:

• Industrial maintenance technician : the most sought-after profile, 12.1% of difficult recruitments
• Mechanical and electrical draftsman-designer : 9.4% of difficult recruitments

The case of welders and boilermakers illustrates the extent of the problem. Their recruitment difficulty rate reaches 80.1% according to the same sources. France plans to train 100,000 welders to meet the EPR2 construction sites and naval propulsion. EDF alone estimates its needs at over 10,000 qualified welders over the next decade, according to Energy Knowledge (May 2025). Even the most cautious estimate remains massive on the scale of a single company. This figure provides a scale: it is not about adjusting a few positions, but about rebuilding a skills pipeline. Recruiting alone will not be enough to cover this volume within the timelines set by major industrial programs.

A vacant position is not neutral for a production line. A machine without a qualified operator runs slowly or stops. An order book does not shorten because recruitment is dragging. Resorting to temporary work or subcontracting sometimes fills the gap, at a higher cost and without a sustainable transfer of skills for the site.

A site that anticipates an increase in pace, a new line, or a quality certification stacks two schedules that do not meet. The industrial project schedule is set by a client or a standard. The recruitment schedule, slower, is not. The latter delays the former, without production having a say.

The double effect: retirements and the cost of non-quality

In addition to this lack of new entrants, there is a massive departure of skills in place. According to franceinfo, one million retirements are expected in the French industry by 2030. The phenomenon primarily affects the already most strained professions: according to the BMO 2026 survey by France Travail, the recruitment difficulty rate in metallurgy and maintenance professions exceeds 70% at the national level. Each retirement takes away gestures and reflexes that no one has documented.

This deficit also has a cost on what comes out of the workshop. According to a study by Afnor conducted in December 2023, 80% of industrial companies place their cost of non-quality between 0 and 5% of their revenue, and 17% anticipate an increase in the coming years, compared to only 8% in 2017. The skills deficit is therefore not limited to vacant positions: it also weighs on the quality produced once the position is filled.

Recruiting, mentoring, on-the-job training: why these responses are no longer sufficient

Traditional recruitment faces a shortage of qualified profiles that affects the entire sector at the same time. A maintenance technician position can remain open for several months before a qualified candidate comes forward. Expanding the search area or revising the salary scale does not change the number of trained profiles available in the national market. Some sites end up hiring a less qualified profile, for lack of better options, and postpone the training problem instead of solving it.

Senior-junior mentoring requires the time of an expert who is engaged in their own tasks. It only trains one junior at a time and depends on a declining population of seniors, the same wave of retirements mentioned earlier. For the rarest tasks, this dependency becomes a point of fragility for the entire site. A site that loses its reference trainer loses its ability to transmit knowledge: the skill goes with the person.

On-the-job training has a hidden cost: each repetition of a rare task immobilizes equipment or slows down a pace. The least frequent tasks remain the least documented and the riskiest to learn in real conditions.

The choice that remains open

Recruitment largely escapes the control of an industrial site: the pool of qualified candidates tightens for everyone at the same time, across the same job markets. One variable remains actionable internally: the speed at which an operator already in position becomes autonomous on a task or procedure. Waiting for the ideal profile, or shortening the skill development of those who are already there. The second path depends neither on the labor market nor on a recruitment agency.

Repeat a rare action without mobilizing a senior or stopping a line

On this specific point, mixed or virtual reality simulation changes the equation. It allows an operator to repeat a rare action as many times as necessary, without immobilizing a machine or monopolizing a senior expert. According to the Swiss HEFP and the Canopé Network, active repetition and interactivity in a simulated environment enhance procedural memory, better than passive observation of a colleague at work. For long procedures, maintenance, or production, this gap weighs on the time required before an operator becomes autonomous. It also weighs on the number of times a senior must interrupt their own work to supervise.

This approach does not replace all actions. It is relevant for repeatable and documentable procedures: a sequence of steps, a control point, a safety sequence. It is less relevant for actions that depend on fine sensory judgment, such as assessing the tension of a material by touch or detecting a sound anomaly on a machine. It is also less relevant in the face of a total unforeseen event, outside of any planned scenario. For those actions, the experience accumulated in the field remains the best training.

What a shorter skill development changes

On the training use cases deployed by Myxed, the time to skill up decreases by 60%. The method:

• Frame the action to be transmitted with the field teams
• Prototype a first testable scenario in a few weeks
• Build the application with the remaining subject matter experts
• Deploy on the relevant position, with a follow-up on the results

Let's take a typical case, for illustration: a quality control position in aerospace, where the inspection action is rare, precise, and where the senior expert who masters it is only available a few hours a week. A simulated scenario allows a new operator to repeat this inspection action as many times as necessary, without monopolizing this expert or interrupting the control line. The number of operators capable of performing this control autonomously increases, without multiplying the hours spent on the senior expert. It is in this type of position, rare and dependent on a single expert, that the gap in skill development time widens the most.

The simulated scenario is replayed identically for each new operator, and is measured: a traceability that informal observation in the field does not provide.